Feasibility assessment of establishing long-term landslide volume inventory in mountainous area of Taiwan: an example of Hsiaolin village

Abstract

Global warming have contributed to the increase of extreme weather and climate events Over the past 40 years the annual rainfall days in Taiwan have decreased and the annual average rainfall intensity has increased The increase of extreme rainfall events will lead to more frequent landslides and debris flows increased soil erosion rates and accelerated changes in surface geomorphology Therefore in order to obtain long-term changes in surface geomorphology large-scale and long-term continuous sources of information are needed FORMOSAT-2 is a daily revisited satellite Since its launch in May 2004 it has accumulated 12 years of observations for Taiwan Using its stereo image to produce digital terrain models (DTM) it can obtain a wide range of long-term surface geomorphology change information For the sake of obtaining more ground information FORMOSAT-2 is less likely to sense with a large viewing angle and it is not easy to obtain a stereo pair with a B/H ratio of 0 6 to 1 that is often used This study uses FORMOSAT-2 narrow baseline stereo to produce a digital terrain model to establish long-term surface geomorphic changes in Taiwan It can be divided into three parts: FORMOSAT-2 image geometric correction SIFT method for image matching and use ENVI DTM extraction module produces digital terrain models The study area is Hsiaolin village where large-scale landslide disaster caused by the heavy rainfall during Typhoon Morakot in 2009 This study will separately produce the digital terrain model of the Hsiaolin village area before and after the landslide compare it with the existing digital elevation model and evaluate the accuracy and estimate the large-scale landslide volume in Hsiaolin village The digital terrain model produced by this study can roughly reflect the trend of terrain Compared with the existing digital elevation model the RMSE before the disaster is 17 99 meters the RMSE after the disaster is 26 81 meters and the estimated volume of landslide is about 3397 3x10?7 m3

Date of Award	2019
Original language	English
Supervisor	Cheng-Chien Liu (Supervisor)